N95- 28832 AUTOMATED FIBER PLACEMENT--EVOLUTION AND CURRENT DEMONSTRATIONS Carroll G. Grant, Hercules Aerospace Company Program Manager, NASA ACT Contracts Composite Structures Group Bacchus, Works, Magna, Utah 84044 Vernon M. Benson, Hercules Aerospace Company Manager, Composite Structures Technology Composite Structures Group Bacchus Works, Magna, Utah 84044 ABSTRACT The automated fiber placement process has been in development at Hercules since 1980. Fiber placement is being developed specifically for aircraft and other high performance structural applications. Several major milestones have been achieved during process development. These milestones are discussed in this paper. The automated fiber placement process is currently being demonstrated on the NASA ACT program. All demonstration projects to date have focused on fiber placement of transport aircraft fuselage structures. Hercules has worked closely with Boeing and Douglas on these demonstration projects. This paper gives a description of demonstration projects and results achieved. CONFERENCE Third NASA Advanced Composites Technology Conference 8- 11 June 1992, Long Beach, California AUTOMATED FIBER PLACEMENT EVOLUTION Hercules filament winding experience dates back to 1957. We have used this experience to develop a new automated process to accurately place and compact prepreg tow material that meets the requirements for aircraft applications. The automated fiber placement process (also known as automated tow placement) has been significantly improved over the last 10 years and is now production ready. 625 PREGEDll_G Ph3Z _L_hiK NOT PJLMED In the late 1970s, Hercules identified composite aircraft and space structures as possible applications for automated filament winding. Although filament winding was ideal for rocket motor cases, it was not necessarily a good process for aircraft and space structures. These parts were seldom symmetrical bodies of revolution; many had complex compound curvatures with concave surfaces. Filament winding relied on tension and geodesic paths to keep the material in place and was limited in ply angle orientation capability. On complex structures, filament winding was not practical. Automated tape layup was limited to relatively flat surfaces, and the process was very immature at the time. A machine was needed that would lay material down at any orientation from 0 to 90°,and that could handle both symmetrical bodies of revolution and complex contoured surfaces. Structures fabricated on this machine needed to be equivalent in performance properties to structures fabricated with prepreg tape and fabric materials. It appeared that the machine would be a hybrid of filament winding, tape laying, and some new technology. In 1980, the design of a six-axis machine that could follow complex contoured surfaces while placing and compacting material directly on the surface was started by Hercules engineers at the Clearfield, Utah filament winding facility. It was originally called "advanced filament winding" or "six-axis filament winding." Two years later, the design was complete and procurement of machine elements was under way. In 1983, Hercules assembled the first six-axis machine. This machine incorporated a three-roll wrist (designed for the robotics industry) along with a horizontal profile machine modeled after the state-of-the-art filament winding machines designed and built by Hercules for in-house use. The delivery system consisted of a standard dry fiber creel, a series of redirects, a hot melt resin impregnation station, and a sophisticated delivery head. The delivery head was capable of delivering 12 tows of material. It was designed to cut one tow at a time, and add one tow at atime when commanded by the software. The cutter and the adder were indexed by two small stepper motors. The tows came in on one level, but were spaced approximately one tow width apart to allow clearance for the cutter and adder for each tow. The tows were converged back together as the tray cavities came closer together. The final delivery roller consisted of a segmented roller, with each segment capable of compacting two tows at a time. These roller segments floated individually to provide compaction of contoured surfaces, while allowing individual fiber speeds across the band. The new machine and process were distinguished from standard filament winding by being designated "fiber placement," because the individual tows of fiber were now being placed precisely on the surface of the part and compacted in place as they were applied. Another name often used today is "automated tow placement." The machine and process described above were used to fabricate the Sikorsky ACAP taiicone (Figure 1). Although the ACAP part was successfully fabricated, a 626 OF,IG,,4,v- F;".f 8LACK AND WHITE PHO IOGI_AP'_ Figure 1. The six-axis automated tow placement machine was used to fabricate the Sikorsky ACAP tail cone. combination of filament winding and fiber placement methods was used. A need to improve the technology in several areas became apparent because the operation did not proceed as smoothly as the theory indicated. However, it was a significant step in the evolution of fiber placement. The software required to program the six-axis machine was developed by Dr. Russ Wilhelmsen, a mathematician and computer scientist with a special ability to manipulate spatial geometries. Development and refinement of an off-line programming system for fiber placement has progressed significantly over the past several years. From the early machine and delivery system to the production ready system available today, there have been many improvements (Figure 2). The more significant evolutionary steps are outlined in the following paragraphs. MACHINE DEVELOPMENT MILESTONES Prepreg Tow Impregnation of dry fiber with a hot melt resin delivery system on the machine had some definite cost advantages. However, resin content control was a problem because the process featured a variable rate with stop and go inconsistencies. Prepreg tow was ordered from material suppliers starting in 1985. The first materials 627 Figure 2. The seven-axis automated tow placement machine incorporates many improvements over earlier machines and delivery systems. used were made from solvent-based impregnation systems. The materials worked well as long as the residual solvent level was well controlled. Too much solvent made the material very sticky and soft. Mechanical properties testing on the Large Fuselage MANTECH Program showed how difficult it was to get all the residual solvent out of the tow during final part cure. This resulted in lower Tg properties for parts made using solvated prepreg tow. To improve Tg properties, the Hercules Materials group developed a hot melt prepreg tow process for the Large Fuselage Program (later known as the V-22 Aft Fuselage Demonstration). Since then, the Hercules hot melt prepreg tow process has been greatly refined. Today's process delivers a well-controlled material form for fiber placement. The cost of prepreg tow today is slightly higher than prepreg tape because of the low volume in use. It will become a lower cost material form than prepreg tape as the volume increases. Bi-Directional Tensioners One of the key elements in fiber placement is the maintenance of a low, consistent tension. The early delivery creels were capable of applying tension, but did not have the ability to respond to a slack fiber condition, which occurs regularly in fiber placement as a result of the wrist motion on complex surfaces. Simple mechanical devices were used on fiber placement machine No. 1(FPM1) to keep fibers tight. Fiber placement machine No. 2 (FPM2) was equipped with tensioners 628 that would allow material to pay out as needed and would take up material on the spool if it attempted to go slack, all the time maintaining constant tension on each tow. Refrigerated Creel Prepreg tow does not use backing paper and is spooled on a way wound package to facilitate removal. Some materials can be spooled at room temperature and unspooled successfully, but the majority of them will not. Hercules added refrigeration to the fiber delivery creels to prolong the life of the material, to allow for clean unspooling, and to protect the material from slump when it was not rotating. This has reduced the problems in the creel to near zero. Ribbonization Thermoset prepreg tow is in a soft pliable material form. When it is manufactured, the supplier tries to control the width and thickness profile of each tow. Typically, a width control of -I-0.025 inches can be guaranteed by the prepreg supplier for a way wound package. After the spools are loaded onto the fiber placement delivery creel, the fiber must pass through several redirects before it enters the delivery head. Depending on the severity of the fiber path (based on wrist position for a given geometry), the soft pliable tow will often change shape slightly while traveling this path. With a stringent requirement to deliver the individual tows onto a structure with no more than a O.030-inch gap or overlap between tows or between bands of tows, the tow width variations just described could easily exceed the gap/overlap requirements. To avoid the inconsistencies described, Hercules developed a technique to ribbonize (control the width and thickness of each tow) within the delivery head. This has allowed us to deliver a wide variety of materials and to accommodate last minute design changes from customers. This ribbonizing module can be easily removed from the head if it is not required. Two Tier Delivery To accommodate individual tow cut and add mechanisms, the tows in the delivery head are separated into two tiers with a one-tow width separation between each tow. The two tiers of tows are merged together near the delivery point. This allows a straight tow path through the head. The individual actuators for each tow allow any tow or combination of tows to be cut or added back in simultaneously. 629 Heavy Duty Wrist A new roll-bend-roll wrist was designed and put into use on FPM2 to accommodate the high compaction forces and newer generation delivery heads. This new wrist was a Hercules design and incorporated a series of compact motor and non-backlash gear assemblies into a wrist package that maintain a high degree of flexibility on each of the three axes. Large Crossfeed Travel For large parts with severe cross section changes and for parts where material must be placed on the end of the part near the shaft or rotational centerline, a flexible wrist and a large crossfeed travel are required. Hercules designed FPM2 with 8.5 feet of crossfeed travel, and 1.5 feet of that travel beyond the spindle centerline. This feature was very helpful in fabricating the V-22 aft fuselage, a demonstration boat hull, and a 4-foot diameter sphere. It is also useful to meet production schedules that require the manufacture of widely different part geometries. Synchronization To have a tow start and stop accurately on the surface of a part required the development of synchronized motion between the rate at which new tows were being added and the rate at which material is being applied to the surface of the part. It also required look aheads in the software. These features were accommodated in later generations of the delivery head. Heating and Cooling Zones Over years of making parts, Hercules learned that strategic heating and cooling of the tows as they are delivered aided in the effective processing of the materials. The tows are cooled to reduce tack and to stiffen them so that they can be fed or pushed. The tows are heated in the ribbonizer to condition their width/thickness control, and are heated slightly at the laydown point to increase the tack characteristics (resin dependent). 630 Direct CATIA and IGES CAD Link to Off-Line Programming To ensure that the complex surfaces being programmed are identical to the ones used in the product and tool design of the structures, Hercules developed direct transfer links from CATIA and IGES to feed three-dimensional CAD data to the fiber placement machine's off-line programming system. This direct link allows a rapid, accurate transfer of surface data. Simulation Software The need for the ability to estimate manufacturing times before the parts are actually built became apparent early in machine development. Hercules developed software to simulate the actual manufacturing times based on a given part geometry and the kinematics of the machine in which it isto be processed. This is a great help in planning work schedules, estimating costs, and in evaluating design changes to the machine. Hercules also has graphic simulation of the machine applying the material to the structure that can be used to verify that everything is in order before part fabrication. Data Logging The machine control system software has been customized over the years to provide avariety of useful data for the fiber placement operator and engineer. Some of these data include: laydown rates, off part times, machine down times, time stamps, system diagnostics, ply and circuit data, time to complete, etc. These are only some of the significant areas where the evolution of the fiber placement machine technology at Hercules has been apparent. The other measure of the advances in technology can be seen by the quality and types of parts that are being fabricated today. The NASA ACT Program is a good illustration of some of the current work. CURRENT DEMONSTRATIONS Hercules ACT Program The Hercules NASA ACT Program was established to demonstrate and validate the low cost potential of the automated tow placement process for fabrication of commercial aircraft primary structures. It iscurrently being conducted as a cooperative program in collaboration with the Boeing ATCAS program. Hercules is 631 responsible for fabrication of test panels that are representative in design of commercial aircraft fuselage Section 46 crown, keel, and side quadrants. Boeing is responsible for panel design and testing (Table I). All panels are fabricated using the automated tow placement process. Table I. Test Matrix for Boeing/Hercules ACT Program Integration Bi- Comp/ Fuselage Undam- Tension Shear Comp. Tension Shear Quad- aged with with with with with rant Test Article Elements Damage Damage Damage Damage Damage Crown Flat, unstiffened 2 skin panels, 60 inchx 150 inch Crown Flat, stiffened _anels, 63 inchx 150 inch Crown Curved, stiffened :_anels, 65 inchx72 inch Keel Flat, coupons, 35 inchx60 inch Keel Flat, stiffened !panels, 30 inchx 44 inch Keel Curved, stiffened _anels, 30 inch x 44 inch Window Tension coupons belt with thick taper 35 inchx60 inch Window Curved panel with belt taper and cutout, 40 inchx 40 inch Window Panel with double belt window frame 40 inchx 40 inch Program activities to date have focused on the fuselage crown quadrant. The Hercules program includes four large test panels for the crown task. Stiffened, unstiffened, flat, and curved panels have been fabricated for tension testing. A hybrid material form was used on two of the panels and the other two panels were made with all graphite materials. The hybrid material consisted of 25% S2 glass and 75% AS4 6K fiber. The glass and graphite tows were impregnated with Fiberite 938 resin. The glass/graphite hybrid material form is produced on the tow placement machine in a 24-tow band with repeat units of two tows of glass and six tows of graphite. The all graphite material form was AS4 6K fiber impregnated with Fiberite 938 resin. 632 The flat unstiffened panels were both 63 inchx 150 inch 15-plytension fracture panels. One panel was 100% graphite and the other was made with the glass/ graphite hybrid material form. Both panels were autoclave cured using a 350°F and 100 psi cure cycle. Both panels looked good, and no defects were found with NDI. These panels were fabricated in July 1991 and were delivered to Boeing for testing. Both panels were tested and results were excellent when compared to prepreg tape layup panels. The 63-inch x 150-inch flat stiffened hybrid panel was cured and delivered in early 1992. This panel was stiffened with five full-length 16-ply hat stringers that were co-cured to the 15-ply hybrid skin (Figure 3). The hat stringers were also made Stringer ply layup [+--45102/90/02/-t- 15/02/90102/-t- 45] _ < Material: 25% S2 glass/75% AS4/938 _ in. typical Repeat unit: 2tows $2 glass 11_, 6 tows A S4 __'_,,_ 11 ___ 14 in. typical -"1 -,--1.25 in. 1. t I 2.2 in. I-,-1.3 in. Skin layu I-,_-4.8 in. , I [_+45/90/0/_+ 601 + 15/90/- 151 ___60101901 -t-45] Figure 3. Hercules NASA ACT Program flat stiffened panel had five full-length, 16-ply hat stringers co-cured to the 1S-ply hybrid skin. with the glass/graphite hybrid material form. The hat stringers were kitted from a flat tow-placed hybrid panel and were hot drape formed. The formed stringers were trimmed to size and fitted with a molded silicon rubber cure mandrel. The stringer/ cure mandrel assemblies were located to the panel inside mold line (IML) (Figure 4). A layer of film adhesive was laid up under the skin flanges of each hat stringer. A peel ply was used on the panel outside mold line (OML) and IMLsurface. After the stringers were located, a molded graphite flex caul was installed on the IML of the panel assembly. The assembled panel was then vacuum bagged and autoclave cured at 350°F. After cure, the panel was trimmed to size, and the molded silicon rubber stringer cure mandrels were removed without difficulty. Quality of the panel was good, and no defects were found with NDI (Figure 5). Boeing will test this panel for axial damage tolerance. 633 Figure 4. The hybrid panel/skin/stringer assembly was located to the panel IML. Figure 5. The cured hybrid panel quality was good; no defects were found with NDI. 634